CN102497556B - A kind of scene change detection method, apparatus, equipment based on time-variation-degree - Google Patents

Abstract

The present invention discloses a kind of scene change detection method, apparatus, equipment based on time-variation-degree.Belong to coding and decoding video field, methods described is by obtaining the y of two field picture to be detected, u, the time-variation-degree statistical property of v information, the different situations changed according to brightness and/or chrominance information, the careful decision threshold of brightness and the careful decision threshold of colourity are determined, according to the change of the statistical information of brightness and/or colourity, the careful decision threshold of brightness and the careful decision threshold of colourity, carries out determining whether that scene switches.It may be such that encoder is adapted dynamically coding strategy using the inventive method, so as to Optimized Coding Based performance.

Description

A kind of scene change detection method, apparatus, equipment based on time-variation-degree

Technical field

The present invention relates to coding and decoding video field, more particularly to a kind of scene change detection side based on time-variation-degree Method, device, equipment.

Background technology

The yuv forms that sources of encoded information separates using brightness with chromatic component, simultaneously because actual coding film source Diversity so that rely solely on the effect that the detection method of brightness or chrominance information can not obtain.And utilize coding information Scene be switched and determined method, can have preferable effect in transcoding or decoding, but compile to acquired original film source The coding side of code, due to lacking the coding information of priori so that be switched and determined method using the scene of coding information and be not suitable for Coding side.Simultaneously, it is contemplated that the complexity of actual film source, that is, the more sufficient film source of chrominance information being present, there is also chrominance information The film source of shortage is (such as：The film source of the camera collection of black and white scene, poor-performing in film film source), corresponding numeric distribution Section is also entirely different, is not added with the unified judgment model distinguished, the accuracy for causing to judge drastically is declined.

The content of the invention

The purpose of the embodiment of the present invention is to propose a kind of scene change detection method based on time-variation-degree, it is intended to solves Encoding efficiency that the detection method of brightness or chrominance information can not obtain certainly is relied solely in the prior art and using compiling The scene of code information is switched and determined method and is not suitable for coding side problem, and is not separately provided the more sufficient piece of chrominance information The problem of source is switched and determined model with the film source scene that chrominance information lacks, and false determination ratio is higher.

The embodiment of the present invention is achieved in that a kind of scene change detection method based on time-variation-degree, its feature It is, methods described includes：

Step A, the y of frame to be detected (t frames) image, u, the time-variation-degree statistical property of v information are obtained respectively；

Step B, the different situations changed according to brightness and/or chrominance information, determines the careful decision threshold of brightness and colourity Careful decision threshold；

Step C, according to the change of the statistical information of brightness and/or colourity, the careful decision threshold of brightness and the careful judgement of colourity Threshold value, carry out determining whether that scene switches.

The step A is specifically included：

Step a, determine frame to be detected (t frames) image key area Regicn_t；

Step b, judges whether the current block in frame to be detected belongs to frame to be detected (t frames) image key area, is then Into step c, otherwise into next piece of block of current block_t,n+1, return to step b and judged；

Step c, calculate current block block_t,nY, u, v information time change degree the first statistical property vector, second system Count eigen vector；

Step d, judge that whether all blocks have all asked for statistical property vector in current frame image key area, are Then enter step e, otherwise into next piece of block of current block_t,n+1, reenter step b；

Step e, the first statistical property, the second statistics for calculating y, u, v information time change degree of two field picture to be detected are special Property；

Wherein：T represents the frame number of frame to be detected in the video sequence, and t frames are frame to be detected,

Region_tT two field picture key areas are represented,

block_t,nN-th piece of t two field pictures is represented, n-th piece is to be detected piece, namely current block,

block_t,n+1Represent (n+1)th piece of t two field pictures.

Y represents the luminance component of image, and u, v represent the chromatic component of image respectively.

The step c is specially：

c1：Obtain block_t,nF information time change degrees 3 set b_f,t-2,n、b_f,t-1,n、b_f,t,n；

F is respectively equal to y, u, v, and y represents the luminance component of image, and u, v represent the chromatic component of image respectively, gathers b_f,t-2,n、b_f,t-1,n、b_f,t,nCalculation formula such as shown in (1)：

M in formula (1) is respectively equal to t-2, t-1, t, you can obtains b respectively_f,t-2,n、b_f,t-1,n、b_f,t,n,

block_m,nN-th piece of m two field pictures are represented,

block_m+1,nN-th piece of m+1 two field pictures are represented,

f_m(i, j) represents the numerical value of m two field picture the i-th row jth row f information,

f_m+1,n(i, j) represents the numerical value of m+1 two field picture the i-th row jth row f information,

f_m(i,j)∈block_m,nExpression is located at block_m,nThe numerical value of m two field pictures the i-th row jth row f information in block,

f_m+1(i,j)∈block_m+1,nExpression is located at block_m+1,nM+1 two field pictures the i-th row jth row f information in block Numerical value,

Expression meets f_m+1(i,j)∈block_m+1,nAnd f_m(i,j)∈block_m,nInstitute Have

f_m+1(i,j)-f_mThe set of (i, j),

f_m+1(i,j)-f_m(i, j) is the subtraction of the numerical value of corresponding f information,

c2：Ask for block_t,n3 set b of f information times change degree_f,t-2,n、b_f,t-1,n、b_f,t,nThe first statistics it is special Property, Std (b are designated as respectively_f,t-2,n)、Std(b_f,t-1,n)、Std(b_f,t,n), Std represents to seek mean square deviation；

c3：Ask for block_t,n3 set b of f information times change degree_f,t-2,n、b_f,t-1,n、b_f,t,nThe second statistics it is special Property, mean (b are designated as respectively_f,t-2,n)、mean(b_f,t-1,n)、mean(b_f,t,n), mean represents to average；

c4：Build block_t,nF information times change degree the first statistical property vector T s_b_f,t,nWith the second statistical property Vector T m_b_f,t,n, its construction method is as follows：

Ts_b_f,t,n=(Std (b_f,t-2,n),Std(b_f,t-1,n),Std(b_f,t,n))(2)

Tm_b_f,t,n=(mean (b_f,t-2,n),mean(b_f,t-1,n),mean(b_f,t,n))(3)

The step e is specially：

To all pieces in t two field picture key areas of Ts_b_f,t,nAverage, the f information times as t two field pictures change First statistical property Ts_frame of degree_f,t, to all pieces in t two field picture key areas of Tm_b_f,t,nAverage, as t Second statistical property Tm_frame of the f information time change degrees of two field picture_f,t,

Ts_frame _f,t=mean (Ts_b_f,t,n)(4)

Tm_frame _f,t=mean (Tm_b_f,t,n)(5)。

The another object of the embodiment of the present invention is to propose a kind of scene change detection device based on time-variation-degree, institute Stating device includes：The time-variation-degree statistical property acquisition module 41 of two field picture y u v information, careful decision threshold acquisition module 42nd, scene is switched and determined module 43,

The time-variation-degree statistical property acquisition module 41 of two field picture y u v information, for obtaining frame to be detected (respectively T frames) image f information time-variation-degree statistical property, f is respectively equal to y, u, v, and y represents the luminance component of image, u, v points Not Biao Shi image chromatic component；

Careful decision threshold acquisition module 42, for the different situations according to brightness and/or chrominance information change, determine bright Spend careful decision threshold and the careful decision threshold of colourity；

Scene is switched and determined module 43, for according to the change of the statistical information of brightness and/or colourity, the careful decision threshold of brightness Value and the careful decision threshold of colourity, determine whether that scene switches.

The another object of the embodiment of the present invention is to propose a kind of scene change detection dress based on time-variation-degree The equipment put.

Beneficial effects of the present invention

The embodiment of the present invention is by obtaining the y of two field picture to be detected, u, the time-variation-degree statistical property of v information, according to Brightness and/or the different situations of chrominance information change, determine the careful decision threshold of brightness and the careful decision threshold of colourity, according to bright The change of the statistical information of degree and/or colourity, the careful decision threshold of brightness and the careful decision threshold of colourity, carry out determining whether scene Switching.It may be such that encoder is adapted dynamically coding strategy using present invention method, so as to Optimized Coding Based performance.

Brief description of the drawings

Fig. 1 is a kind of scene change detection method flow diagram based on time-variation-degree of the embodiment of the present invention；

Fig. 2 is the method detailed flow chart of step S11 in Fig. 1；

Fig. 3 is step S113 method detailed flow charts in Fig. 2；

Fig. 4 is a kind of scene change detection structure drawing of device based on time-variation-degree of the embodiment of the present invention；

Fig. 5 is the structure chart of the frame module 41 in Fig. 4 devices.

Embodiment

In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and examples The present invention is further elaborated, and for convenience of description, illustrate only the part related to the embodiment of the present invention.It should manage Solution, the specific embodiment that this place is described, it is used only for explaining the present invention, is not intended to limit the invention.

The embodiment of the present invention is by obtaining the y of two field picture to be detected, u, the time-variation-degree statistical property of v information, according to Brightness and/or the different situations of chrominance information change, determine the careful decision threshold of brightness and the careful decision threshold of colourity, according to bright The change of the statistical information of degree and/or colourity, the careful decision threshold of brightness and the careful decision threshold of colourity, carry out determining whether scene Switching.It may be such that encoder is adapted dynamically coding strategy using present invention method, so as to Optimized Coding Based performance.

Fig. 1 is a kind of scene change detection method flow diagram based on time-variation-degree of the embodiment of the present invention, the side Method comprises the following steps：

S11：The y of frame to be detected (t frames) image, u, the time-variation-degree statistical property of v information are obtained respectively；

Fig. 2 is the y of the acquisition two field picture to be detected of the embodiment of the present invention, u, the time-variation-degree statistical property side of v information Method flow chart；

Described " obtaining the y of frame to be detected (t frames) image, u, the time-variation-degree statistical property of v information respectively " is specific Comprise the following steps：

S111：Frame to be detected (t frames) image key area is determined, is designated as Region_t；

T represents the frame number of frame to be detected in the video sequence, in embodiments of the present invention, if t frames are frame to be detected； Region_tRepresent t two field picture key areas；

Key area Region that can be using whole t two field pictures as t two field pictures_t；

The central area of t two field pictures can also be taken as t two field picture key areas：It is most upper such as to remove t two field pictures The height/k at end_hIndividual macro-block line, the height/k of bottom_hIndividual macro-block line, the width/ for removing the t two field pictures leftmost side again k_wIndividual macro block row, rightmost side width/k_wAfter individual macro-block line, remaining image is as Region_t。

Wherein height, width is respectively the line number of image pixel, columns, k_h、k_wRespectively line direction and column direction ratio Example coefficient, optional 16,8 etc. integers；The block of typically small proportionality coefficient is adapted to the weaker film source of texture information；

In the case where ensureing certain accuracy of detection, the key area of image can be determined by reducing determinating area, so as to reduce Amount of calculation, specifically preferably result can be obtained to count by testing, the accuracy of detection is set according to being actually needed.

Further, can also be by carrying out down-sampling to original image to reduce amount of calculation, now, before step S11 also Including step：Down-sampling is carried out to pending original image；

The Downsapling method that the method for down-sampling includes commonly using in field of the present invention is carried out to original image, is the art Common knowledge, in this not specified (NS).

S112：Judge the current block block in frame to be detected_t,nWhether to be detected frame (t frame) image key area is belonged to Domain, i.e. block_t,n∈Region_t, it is then to enter step S113, otherwise into next piece of block of current block_t,n+1, return to Step S112 is judged；

T represents the frame number of frame to be detected in the video sequence, in embodiments of the present invention, if t

Frame is frame to be detected, namely present frame；

block_t,nN-th piece of t two field pictures is represented, in embodiments of the present invention, if n-th piece is to be detected piece, namely is worked as Preceding piece；

block_t,n+1Represent (n+1)th piece of t two field pictures；

Wherein the size of block can adjust as needed, conventional such as 16x16,8x8,32x32；

S113, calculate current block block_t,nY, u, v information time change degree the first statistical property vector, second system Count eigen vector.

If f is respectively equal to y, u, v, y represents the luminance component of image, and u, v represent the chromatic component of image respectively, then " meter Calculate current block block_t,nY, u, the time-variation-degree statistical property of v information " namely " calculate current block block_t,nF information Time-variation-degree statistical property ".

It is described " to calculate current block block_t,nF information time-variation-degree statistical property " comprise the following steps:Fig. 3 is Step S113 method detailed flow charts in Fig. 2；

S1131：Obtain block_t,nF information time change degrees 3 set b_f,t-2,n、b_f,t-1,n、b_f,t,n；

The calculation formula of set is such as shown in (1)：Even the m in formula (1) is respectively equal to t-2, t-1, t, you can obtains respectively Take b_f,t-2,n、b_f,t-1,n、b_f,t,n。

block_m,nRepresent n-th piece of m two field pictures；The optional conventional macroblock size of the size of block in the embodiment of the present invention 16x16, either larger block such as 32x32 or less piece such as 8x8；Generally large block is adapted to the big coded slice of resolution ratio The weaker film source of source, texture information；Typically small block is adapted to the small coding film source of resolution ratio；

block_m+1,nRepresent n-th piece of m+1 two field pictures；

f_m(i, j) represents the numerical value of m two field picture the i-th row jth row f information；

f_m+1,n(i, j) represents the numerical value of m+1 two field picture the i-th row jth row f information；

f_m(i,j)∈block_m,nExpression is located at block_m,nThe numerical value of m two field pictures the i-th row jth row f information in block；

f_m+1(i,j)∈block_m+1,nExpression is located at block_m+1,nThe i-th row of m+1 two field pictures jth row in block

The numerical value of f information；

Expression meets f_m+1(i,j)∈block_m+1,nAnd f_m(i,j)∈block_m,nInstitute Have

f_m+1(i,j)-f_mThe set of (i, j)；

f_m+1(i,j)-f_m(i, j) is the subtraction of the numerical value of corresponding f information；

S1132：Ask for block_t,n3 set b of f information times change degree_f,t-2,n、b_f,t-1,n、b_f,t,nFirst statistics Characteristic, Std (b are designated as respectively_f,t-2,n)、Std(b_f,t-1,n)、Std(b_f,t,n)；Std represents to seek mean square deviation；

S1133：Ask for block_t,n3 set b of f information times change degree_f,t-2,n、b_f,t-1,n、b_f,t,nSecond statistics Characteristic, mean (b are designated as respectively_f,t-2,n)、mean(b_f,t-1,n)、mean(b_f,t,n)；Mean represents to average；

S1134：Build block_t,nThe second statistical property of f information times change degree the first statistical property vector sum vector, Ts_b is designated as respectively_f,t,n, Tm_b_f,t,n, its construction method is as follows：

Ts_b_f,t,n=(Std (b_f,t-2,n),Std(b_f,t-1,n),Std(b_f,t,n))(2)

Tm_b_f,t,n=(mean (b_f,t-2,n),mean(b_f,t-1,n),mean(b_f,t,n))(3)

In the embodiment of the present invention, the component positions of institute's directed quantity are not fixed, and can arbitrarily be converted, and are represented to simplify, The expression formula of institute's directed quantity all only lists a kind of spread pattern of component in the embodiment of the present invention；

In the embodiment of the present invention, all computings carried out to vector are vector operations, i.e., each component of vector are distinguished Computing is carried out, operation result is still vector；

In the embodiment of the present invention, in order to simplify expression, vector is represented using sequence number of the component of vector in vector Each component,

Such as Ts_b_f,t,nEach component be represented by：

Ts_b _f,t,n(1)=Std (b_f,t-2,n)；Ts_b_f,t,n(2)=Std (b_f,t-1,n)；Ts_b_f,t,n(3)=Std (b_f,t,n)；

Such as Tm_b_f,t,nEach component be represented by：

Tm_b _f,t,n(1)=mean (b_f,t-2,n)；Tm_b_f,t,n(2)=mean (b_f,t-1,n)；Tm_b_f,t,n(3)=mean (b_f,t,n)。

S114：Judge whether all blocks have all asked for statistical property vector in t two field picture key areas, be to enter Enter step S105, otherwise into next piece of block of current block_t,n+1, reenter step S112；

S115, the first statistical property, the second statistics for calculating y, u, v information time change degree of two field picture to be detected are special Property.

Specially：To all pieces in t two field picture key areas of Ts_b_f,t,nAverage, the f information as t two field pictures First statistical property Ts_frame of time-variation-degree_f,t；To all pieces in t two field picture key areas of Tm_b_f,t,nAverage, The second statistical property Tm_frame as the f information time change degrees of t two field pictures_f,t；

Ts_frame _f,t=mean (Ts_b_f,t,n)(4)

Tm_frame _f,t=mean (Tm_b_f,t,n)(5)

(2), (3) are utilized further to be embodied as (4), (5) above formula

Further, in order to simplify expression, vector each is represented using sequence number of the component of vector in vector Component, such as

Ts_frame_f,tEach component be represented by：

Tm_frame_f,tEach component be represented by：

Make f in (6) be respectively equal to y, u, v, ask for Ts_frame_y,t, Ts_frame_u,t, Ts_frame_v,t,

Make f in (7) be respectively equal to y, u, v, ask for Tm_frame_y,t, Tm_frame_u,t, Tm_frame_v,t；

Ts_frame_y,tReferred to as the first statistical property of t two field pictures monochrome information time-variation-degree；

Ts_frame_u,t, Ts_frame_v,tReferred to as the first statistical property of t two field pictures chrominance information time-variation-degree；

Tm_frame_y,tReferred to as the second statistical property of t two field pictures monochrome information time-variation-degree；

Tm_frame_u,t, Tm_frame_v,tReferred to as the second statistical property of t two field pictures chrominance information time-variation-degree；

Further, in order to simplify expression, vector each is represented using sequence number of the component of vector in vector Component.

F in (8) is made to be respectively equal to y, u, v；Ts_frame can be obtained_y,t、Ts_frame_u,t、Ts_frame_v,tEach component Representation；

F in (9) is made to be respectively equal to y, u, v, you can to obtain Tm_frame_y,t、Tm_frame_u,t、Tm_frame_v,tEach component Representation；

Ts_frame_y,tEach component be represented by：

Ts_frame _y,t(1)=mean (Std (b_y,t-2,n))；

Ts_frame _y,t(2)=mean (Std (b_y,t-1,n))；

Ts_frame _y,t(3)=mean (Std (b_y,t,n))；

Ts_frame_u,tEach component be represented by：

Ts_frame _u,t(1)=mean (Std (b_u,t-2,n))；

Ts_frame _u,t(2)=mean (Std (b_u,t-1,n))；

Ts_frame _u,t(3)=mean (Std (b_u,t,n))；

Ts_frame_v,tEach component be represented by：

Ts_frame _v,t(1)=mean (Std (b_v,t-2,n))；

Ts_frame _v,t(2)=mean (Std (b_v,t-1,n))；

Ts_frame _v,t(3)=mean (Std (b_v,t,n))；

Tm_frame_y,tEach component be represented by：

Tm_frame _y,t(1)=mean (mean (b_y,t-2,n))；

Tm_frame _y,t(2)=mean (mean (b_y,t-1,n))；

Tm_frame _y,t(3)=mean (mean (b_y,t,n))；

Tm_frame_u,tEach component be represented by：

Tm_frame _u,t(1)=mean (mean (b_u,t-2,n))；

Tm_frame _u,t(2)=mean (mean (b_u,t-1,n))；

Tm_frame _u,t(3)=mean (mean (b_u,t,n))；

Tm_frame_v,tEach component be represented by：

Tm_frame _v,t(1)=mean (mean (b_v,t-2,n))；

Tm_frame _v,t(2)=mean (mean (b_v,t-1,n))；

Tm_frame _v,t(3)=mean (mean (b_v,t,n))；

It will be understood to those skilled in the art that when each formula calculates y, u, v information of two field picture to be detected more than utilizing Between the computational methods of the statistical property of change degree first and the second statistical property be the preferable a kind of calculating side of the embodiment of the present invention Method, other equivalent conventional computational methods are within the scope of the present invention.

S12：The different situations changed according to brightness and/or chrominance information, determine that the careful decision threshold of brightness and colourity are thin Cause decision threshold；

If

((Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)>Thres_u1||

(Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)>Thres_u1)

Or

((Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)>Thres_v1||

(Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)>Thres_v1)

Then：(chrominance information changes violent situation)

It is respectively brightness first kind decision threshold Thres_y_1 to make the careful decision threshold Thres_y1 and Thres_y2 of brightness And Thres_y_2；The careful decision threshold Thres_uv1 and Thres_uv2 of assumed appearance degree is respectively the careful decision threshold of the colourity first kind Value Thres_uv_1 and Thres_uv_2；I.e.

Thres_y1=Thres_y_1, Thres_y2=Thres_y_2

Thres_uv1=Thres_uv_1, Thres_uv2=Thres_uv_2

Then, into step S13 careful decision stage

Wherein, Thres_y_1 is the relative threshold of brightness first kind decision threshold, can be obtained by statistical experiment, i.e., Image can be lacked by counting the non-chrominance information of a large amount of (at least 25 scenes switch film source)

(Ts_frame_y,t(2)-Ts_frame_y,t(1))/Ts_frame_y,t(2)、

(Ts_frame_y,t(2)-Ts_frame_y,t(3))/Ts_frame_y,t(2)

Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright first kind decision threshold relative threshold；

Thres_y_2 is brightness first kind decision threshold absolute threshold, can be obtained by statistical experiment, you can to pass through The non-chrominance information of statistics a large amount of (at least 25 scene switching film sources) lacks image Ts_frame_y,t(2)-Ts_frame_y,t (1)、

Ts_frame_y,t(2)-Ts_frame_y,t(3)

Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright first kind decision threshold absolute threshold；

Thres_uv_1 is colourity first kind decision threshold value difference threshold value, can be obtained by statistical experiment, you can to pass through The non-chrominance information of statistics a large amount of (at least 25 scene switching film sources) lacks image fabs (Tm_frame_u,t(2)-Tm_ frame_v,t(2))

Numeric distribution, determine numerical value corresponding to maximum probability as colourity first kind decision threshold value difference threshold value；

Thres_uv_2 is colourity first kind decision threshold and threshold value, can be obtained by statistical experiment, you can to pass through The non-chrominance information of statistics a large amount of (at least 25 scene switching film sources) lacks image Tm_frame_u,t(2)+Tm_frame_v,t(2)

Numeric distribution, determine numerical value corresponding to maximum probability as colourity first kind decision threshold and threshold value；

Thres_u1 is the relative threshold that colourity u is corresponded to when chrominance information changes drastic scene, can pass through statistical experiment Obtain, you can to lack image by the non-chrominance information for counting a large amount of (film sources that at least 25 scenes switch)

(Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)

(Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)

Numeric distribution, determine to correspond to colourity u's when numerical value corresponding to maximum probability changes drastic scene as chrominance information Relative threshold.

Thres_v1 is the relative threshold that colourity v is corresponded to when chrominance information changes drastic scene, can pass through statistical experiment Obtain, you can to lack image by the non-chrominance information for counting a large amount of (film sources that at least 25 scenes switch)

(Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)

(Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)

Numeric distribution, determine to correspond to colourity v's when numerical value corresponding to maximum probability changes drastic scene as chrominance information Relative threshold.Non- chrominance information lacks image and refers to that the colourity energy in image at least in the presence of a pixel is more than decision threshold Thres_upImage

I.e.

WhereinFor the colourity energy of a pixel, u (i, j), v (i, j), Respectively it is located at image the i-th row j row chromatic components u, v numerical value, Thres_upLack the decision threshold of image for non-chrominance information Value, Thres_up>30；

Else if

((Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)>Thres_u2||

(Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)>Thres_u2))

And

((Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)>Thres_v2||

(Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)>Thres_v2)

Then：(chrominance information changes the situation of small scene)

It is respectively brightness the second class decision threshold Thres_y_3 to make the careful decision threshold Thres_y1 and Thres_y2 of brightness And Thres_y_4, the careful decision threshold Thres_uv1 and Thres_uv2 of assumed appearance degree are respectively the careful decision threshold of the class of colourity second Value Thres_uv_3 and Thres_uv_4；I.e.

Thres_y1=Thres_y_3, Thres_y2=Thres_y_4

Thres_uv1=Thres_uv_3, Thres_uv2=Thres_uv_4

Then, into step S13 careful decision stage

Wherein, Thres_y_3 is brightness the second class decision threshold relative threshold, can be obtained by statistical experiment, you can To lack image by the chrominance information for counting a large amount of (at least 25 scenes switch film source)

(Ts_frame_y,t(2)-Ts_frame_y,t(1))/Ts_frame_y,t(2)、

(Ts_frame_y,t(2)-Ts_frame_y,t(3))/Ts_frame_y,t(2)

Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright the second class decision threshold relative threshold；

Thres_y_4 is brightness the second class decision threshold absolute threshold, can be obtained by statistical experiment, you can to pass through The chrominance information of statistics a large amount of (at least 25 scene switching film sources) lacks image Ts_frame_y,t(2)-Ts_frame_y,t(1)、

Ts_frame_y,t(2)-Ts_frame_y,t(3)

Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright the second class decision threshold absolute threshold；

Thres_uv_3 is colourity the second class decision threshold value difference threshold value, can be obtained by statistical experiment, you can to pass through The chrominance information of statistics a large amount of (at least 25 scene switching film sources) lacks image fabs (Tm_frame_u,t(2)-Tm_frame_v,t (2))

Numeric distribution, determine numerical value corresponding to maximum probability as colourity the second class decision threshold value difference threshold value；

Thres_uv_4 is colourity the second class decision threshold and threshold value, can be obtained by statistical experiment, you can to pass through The chrominance information of statistics a large amount of (at least 25 scene switching film sources) lacks image Tm_frame_u,t(2)+Tm_frame_v,t(2)

Numeric distribution, determine numerical value corresponding to maximum probability as colourity the second class decision threshold and threshold value；

Thres_u2 is the relative threshold that colourity u is corresponded to when chrominance information changes small scene, can be obtained by statistical experiment Take, you can to lack image by the chrominance information for counting a large amount of (film sources that at least 25 scenes switch)

(Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)

(Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)

Numeric distribution, determine to correspond to colourity u phase when numerical value corresponding to maximum probability changes small scene as chrominance information To threshold value.

Thres_v2 is the relative threshold that colourity v is corresponded to when chrominance information changes small scene, can be obtained by statistical experiment Take, you can to lack image by the chrominance information for counting a large amount of (film sources that at least 25 scenes switch)

(Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)

(Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)

Numeric distribution, determine that numerical value corresponding to maximum probability changes small scene as chrominance information and corresponds to the relative of colourity v Threshold value.Chrominance information lacks image and refers to that image all pixels point colourity energy is respectively less than decision threshold Thres_downImage

I.e.

WhereinFor the colourity energy of a pixel, u (i, j), v (i, j), divide Wei not be positioned at image the i-th row j row chromatic components u, v numerical value, Thres_downLack the decision threshold of image for chrominance information, Thres_down<15,

Relative threshold Thres_y_1, the brightness first kind decision threshold absolute threshold of the brightness first kind decision threshold Thres_y_2, colourity first kind decision threshold value difference threshold value Thres_uv_1, colourity first kind decision threshold and threshold value Thres_ When colourity u relative threshold Thres_u1, chrominance information change drastic scene is corresponded to when uv_2, chrominance information change drastic scene Corresponding colourity v relative threshold Thres_v1, brightness the second class decision threshold relative threshold Thres_y_3, the class of brightness second are sentenced Determine threshold value absolute threshold Thres_y_4, colourity the second class decision threshold value difference threshold value Thres_uv_3, colourity the second class decision threshold Colourity u relative threshold Thres_u2, chrominance information change is corresponded to when changing small scene with threshold value Thres_uv_4, chrominance information In the relative threshold Thres_v2 that colourity v is corresponded to during small scene, numeric distribution maximum is general used by each threshold value acquisition methods Rate method can also be substituted for averaging method, that is, useAs threshold value, whereinRepresent to sum to k, k represents statistical variable Concrete numerical value, p (k) represents the probability that numerical value k occurs.

Otherwise：

The non-scene switch frame of t frames is determined, makes t=t+1, reenters the judgement that S111 enters next frame.

S13：According to the change of the statistical information of brightness and/or colourity, the careful decision threshold of brightness and the careful decision threshold of colourity Value, carry out determining whether that scene switches.

If

((Ts_frame_y,t(2)-Ts_frame_y,t(1))/Ts_frame_y,t(2)>Thres_y1||

(Ts_frame_y,t(2)-Ts_frame_y,t(3))/Ts_frame_y,t(2)>Thres_y1)

And

((Ts_frame_y,t(2)-Ts_frame_y,t(1))>Thres_y2||

And

(fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))>Thres_uv1||

fabs(Tm_frame_u,t(2)+Tm_frame_v,t(2))>Thres_uv2)

And

(fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))+△>fabs(Tm_frame_u,t(1)-Tm_frame_v,t (1))

&&

fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))+△>fabs(Tm_frame_u,t(3)-Tm_frame_v,t (3)))

Then：Determine t two field pictures F_tFor the start frame of new scene.

Wherein, △ is drift value constant, strengthens the adaptability of algorithm, that is, strengthens simple binaryzation and judge in statistical threshold The problem of causing algorithm performance to decline when bad.△ again may be by statistical experiment acquisition, i.e., by counting largely (extremely The film source of few 25 scenes switching) image

fabs(Tm_frame_u,t(1)-Tm_frame_v,t(1))-fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))

fabs(Tm_frame_u,t(3)-Tm_frame_v,t(3))-fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))

Numeric distribution, determine numerical value corresponding to maximum probability as drift value constant △.Drift value constant △ is asked for be adopted Numeric distribution maximum probabilistic method can also be substituted for averaging method, that is, useAs threshold value, whereinExpression pair K sums, and k represents the concrete numerical value of statistical variable, and p (k) represents the probability that numerical value k occurs.Other conventional statistic laws are applicable.

“||”、“&&”、“fabs”："or", "AND" respectively in C language, " take absolute value computing ".

The embodiment of the present invention is by obtaining the y of two field picture to be detected, u, the time-variation-degree statistical property of v information, according to Brightness and/or the different situations of chrominance information change, determine the careful decision threshold of brightness and the careful decision threshold of colourity, according to bright The change of the statistical information of degree and/or colourity, the careful decision threshold of brightness and the careful decision threshold of colourity, carry out determining whether scene Switching.It may be such that encoder is adapted dynamically coding strategy using present invention method, so as to Optimized Coding Based performance.

Fig. 4 is a kind of scene change detection structure drawing of device based on time-variation-degree of the embodiment of the present invention；The dress Put including：The time-variation-degree statistical property acquisition module 41 of two field picture y u v information, careful decision threshold acquisition module 42, Scene is switched and determined module 43；

The time-variation-degree statistical property acquisition module 41 of two field picture y u v information, for obtaining frame to be detected (respectively T frames) image f information time-variation-degree statistical property, f is respectively equal to y, u, v, and y represents the luminance component of image, u, v points Not Biao Shi image chromatic component；

Careful decision threshold acquisition module 42, for the different situations according to brightness and/or chrominance information change, determine bright Spend careful decision threshold and the careful decision threshold of colourity；

Scene is switched and determined module 43, for according to the change of the statistical information of brightness and/or colourity, the careful decision threshold of brightness Value and the careful decision threshold of colourity, determine whether that scene switches.

Further, the time-variation-degree statistical property acquisition module 41 of the two field picture y u v information also includes：Frame figure As key area acquisition module 411, the first judge module 412, block statistical property vector calculation module 413, the second judge module 414th, the time-variation-degree statistical property computing module 415 of two field picture y, u, v information；

Fig. 5 is the structure chart of the module 41 in Fig. 4 devices.

Two field picture key area acquisition module 411, for determining the image key area of frame to be detected (t frames)；

Whether the first judge module 412, the current block for judging in frame to be detected belong to frame to be detected (t frames) image Key area, it is then to enter module 413, otherwise into next piece of current block, returns to module 412 and judged；T represents to be detected The frame number of frame in the video sequence, if t frames are frame to be detected；

Block statistical property vector calculation module 413, for calculating current block block_t,nY, u, v information time change degree The first statistical property vector, the second statistical property vector；

Second judge module 414, for judging whether all blocks have all been asked for counting in t two field picture key areas Eigen vector, it is then to enter module 415, otherwise into next piece of current block, returns to module 412；

The time-variation-degree statistical property computing module 415 of two field picture y, u, v information, for according to described piece of statistical property Vector, calculate the first statistical property, the second statistical property of y, u, v information time change degree of two field picture to be detected；

Further, the module 413 also includes：The f information time change degree set determining module 4131 of block, the f of block Second system of the f information time change degree set of the first statistical property acquisition module 4132, block of information time change degree set Count the first and second statistical property of f information times change degree vector structure module 4134 of characteristic acquisition module 4133, block.

The f information time change degree set determining module 4131 of block, for obtaining the letters of to be detected piece of f in frame to be detected 3 set of time-variation-degree are ceased, wherein f is respectively equal to y, u, v, and y represents the luminance component of image, and u, v represent image respectively Chromatic component；

The calculation formula of set is such as shown in (1)：Even the m in formula (1) is respectively equal to t-2, t-1, t, you can obtains respectively Take b_f,t-2,n、b_f,t-1,n、b_f,t,n, b_f,t-2,n、b_f,t-1,n、b_f,t,nRespectively to be detected piece of block_t,nF information time change degrees 3 set

block_m,nRepresent n-th piece of m two field pictures；The optional conventional macroblock size of the size of block in the embodiment of the present invention 16x16, either larger block such as 32x32 or less piece such as 8x8；Generally large block is adapted to the big coded slice of resolution ratio The weaker film source of source, texture information；Typically small block is adapted to the small coding film source of resolution ratio；

block_m+1,nRepresent n-th piece of m+1 two field pictures；

f_m(i, j) represents the numerical value of m two field picture the i-th row jth row f information；

f_m+1,n(i, j) represents the numerical value of m+1 two field picture the i-th row jth row f information；

f_m(i,j)∈block_m,nExpression is located at block_m,nThe numerical value of m two field pictures the i-th row jth row f information in block；

f_m+1(i,j)∈block_m+1,nExpression is located at block_m+1,nThe i-th row of m+1 two field pictures jth row in block

The numerical value of f information；

Expression meets f_m+1(i,j)∈block_m+1,nAnd f_m(i,j)∈block_m,nInstitute Have

f_m+1(i,j)-f_mThe set of (i, j)；

f_m+1(i,j)-f_m(i, j) is the subtraction of the numerical value of corresponding f information；

First statistical property acquisition module 4132 of the f information time change degree set of block, for asking in frame to be detected The set of to be detected piece of f information times change degree 3 the first statistical properties；

Specially：Ask for be detected piece of block_t,n3 set b of f information times change degree_f,t-2,n、b_f,t-1,n、b_f,t,n The first statistical property, be designated as Std (b respectively_f,t-2,n)、Std(b_f,t-1,n)、Std(b_f,t,n)；Std represents to seek mean square deviation；

Second statistical property acquisition module 4133 of the f information time change degree set of block, for asking in frame to be detected The set of to be detected piece of f information times change degree 3 the second statistical properties；

Specially：Ask for be detected piece of block_t,n3 set b of f information times change degree_f,t-2,n、b_f,t-1,n、b_f,t,n The second statistical property, be designated as mean (b respectively_f,t-2,n)、mean(b_f,t-1,n)、mean(b_f,t,n)；Mean represents to average；

The first and second statistical property of f information times change degree vector structure module 4134 of block, it is to be detected for building To be detected piece of the second statistical property of f information times change degree the first statistical property vector sum vector in frame；

Specially：Build block_t,nF information times change degree the second statistical property of the first statistical property vector sum to Amount, is designated as Ts_b respectively_f,t,n, Tm_b_f,t,n, its construction method is as follows：

Ts_b_f,t,n=(Std (b_f,t-2,n),Std(b_f,t-1,n),Std(b_f,t,n))(2)

Tm_b_f,t,n=(mean (b_f,t-2,n),mean(b_f,t-1,n),mean(b_f,t,n))(3)

In the embodiment of the present invention, the component positions of institute's directed quantity are not fixed, and can arbitrarily be converted, and are represented to simplify, The expression formula of institute's directed quantity all only lists a kind of spread pattern of component in the embodiment of the present invention；

In the embodiment of the present invention, all computings carried out to vector are vector operations, i.e., each component of vector are distinguished Computing is carried out, operation result is still vector；

In the embodiment of the present invention, in order to simplify expression, vector is represented using sequence number of the component of vector in vector Each component,

Such as Ts_b_f,t,nEach component be represented by：

Ts_b_f,t,n(1)=Std (b_f,t-2,n)；Ts_b_f,t,n(2)=Std (b_f,t-1,n)；Ts_b_f,t,n(3)=Std (b_f,t,n)；

Such as Tm_b_f,t,nEach component be represented by：

Tm_b_f,t,n(1)=mean (b_f,t-2,n)；Tm_b_f,t,n(2)=mean (b_f,t-1,n)；Tm_b_f,t,n(3)=mean (b_f,t,n)。

Further, it is described " to calculate the first system of y, u, v information time change degree of two field picture to be detected in module 415 Count characteristic, the second statistical property " be specially：

To all pieces in t two field picture key areas of f information times change degree the first statistical property vector T s_b_f,t,nAsk equal Value, the first statistical property Ts_frame as the f information time change degrees of t two field pictures_f,t；To t two field picture key areas Interior all pieces of f information times change degree the second statistical property vector T m_b_f,t,nAverage, during f information as t two field pictures Between change degree the second statistical property Tm_frame_f,t；

Ts_frame _f,t=mean (Ts_b_f,t,n)(4)

Tm_frame _f,t=mean (Tm_b_f,t,n)(5)

(2), (3) are utilized further to be embodied as (4), (5) above formula

In order to simplify expression, each component of vector is represented using sequence number of the component of vector in vector, such as

Ts_frame_f,tEach component be represented by：

Tm_frame_f,tEach component be represented by：

Make f in (6) be respectively equal to y, u, v, ask for Ts_frame_y,t, Ts_frame_u,t, Ts_frame_v,t,

Make f in (7) be respectively equal to y, u, v, ask for Tm_frame_y,t, Tm_frame_u,t, Tm_frame_v,t；

Ts_frame_y,tReferred to as the first statistical property of t two field pictures monochrome information time-variation-degree；

Ts_frame_u,t, Ts_frame_v,tReferred to as the first statistical property of t two field pictures chrominance information time-variation-degree；

Tm_frame_y,tReferred to as the second statistical property of t two field pictures monochrome information time-variation-degree；

Tm_frame_u,t, Tm_frame_v,tReferred to as the second statistical property of t two field pictures chrominance information time-variation-degree；

In order to simplify expression, each component of vector is represented using sequence number of the component of vector in vector.

F in (8) is made to be respectively equal to y, u, v；Ts_frame can be obtained_y,t、Ts_frame _u,t、Ts_frame_v,tEach component Representation；

F in (9) is made to be respectively equal to y, u, v, you can to obtain Tm_frame_y,t、Tm_frame_u,t、Tm_frame_v,tEach component Representation；

Ts_frame_y,tEach component be represented by：

Ts_frame _y,t(1)=mean (Std (b_y,t-2,n))；

Ts_frame _y,t(2)=mean (Std (b_y,t-1,n))；

Ts_frame _y,t(3)=mean (Std (b_y,t,n))；

Ts_frame_u,tEach component be represented by：

Ts_frame _u,t(1)=mean (Std (b_u,t-2,n))；

Ts_frame _u,t(2)=mean (Std (b_u,t-1,n))；

Ts_frame _u,t(3)=mean (Std (b_u,t,n))；

Ts_frame_v,tEach component be represented by：

Ts_frame _v,t(1)=mean (Std (b_v,t-2,n))；

Ts_frame _v,t(2)=mean (Std (b_v,t-1,n))；

Ts_frame _v,t(3)=mean (Std (b_v,t,n))；

Tm_frame_y,tEach component be represented by：

Tm_frame _y,t(1)=mean (mean (b_y,t-2,n))；

Tm_frame _y,t(2)=mean (mean (b_y,t-1,n))；

Tm_frame _y,t(3)=mean (mean (b_y,t,n))；

Tm_frame_u,tEach component be represented by：

Tm_frame _u,t(1)=mean (mean (b_u,t-2,n))；

Tm_frame _u,t(2)=mean (mean (b_u,t-1,n))；

Tm_frame _u,t(3)=mean (mean (b_u,t,n))；

Tm_frame_v,tEach component be represented by：

Tm_frame _v,t(1)=mean (mean (b_v,t-2,n))；

Tm_frame _v,t(2)=mean (mean (b_v,t-1,n))；

Tm_frame _v,t(3)=mean (mean (b_v,t,n))；

Further, " it is used for the different situations changed according to brightness and/or chrominance information in module 42, determines that brightness is thin Cause decision threshold and the careful decision threshold of colourity " be specially：

Judge, if

((Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)>Thres_u1||

(Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)>Thres_u1)

Or

((Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)>Thres_v1||

(Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)>Thres_v1)

Then：

It is respectively brightness first kind decision threshold Thres_y_1 to make the careful decision threshold Thres_y1 and Thres_y2 of brightness And Thres_y_22；The careful decision threshold Thres_uv1 and Thres_uv2 of assumed appearance degree is respectively the careful decision threshold of the colourity first kind Value Thres_uv_1 and Thres_uv_2；I.e.

Thres_y1=Thres_y_1, Thres_y2=Thres_y_2

Thres_uv1=Thres_uv_1, Thres_uv2=Thres_uv_2

Into module 43,

Wherein, Thres_y_1 is the relative threshold of brightness first kind decision threshold, can be obtained by statistical experiment, i.e., Image can be lacked by counting the non-chrominance information of a large amount of (at least 25 scenes switch film source)

(Ts_frame_y,t(2)-Ts_frame_y,t(1))/Ts_frame_y,t(2)、

(Ts_frame_y,t(2)-Ts_frame_y,t(3))/Ts_frame_y,t(2)

Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright first kind decision threshold relative threshold；

Thres_y_2 is brightness first kind decision threshold absolute threshold, can be obtained by statistical experiment, you can to pass through The non-chrominance information of statistics a large amount of (at least 25 scene switching film sources) lacks image Ts_frame_y,t(2)-Ts_frame_y,t (1)、

Ts_frame_y,t(2)-Ts_frame_y,t(3)

Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright first kind decision threshold absolute threshold；

Thres_uv_1 is colourity first kind decision threshold value difference threshold value, can be obtained by statistical experiment, you can to pass through The non-chrominance information of statistics a large amount of (at least 25 scene switching film sources) lacks image fabs (Tm_frame_u,t(2)-Tm_ frame_v,t(2))

Numeric distribution, determine numerical value corresponding to maximum probability as colourity first kind decision threshold value difference threshold value；

Thres_uv_2 is colourity first kind decision threshold and threshold value, can be obtained by statistical experiment, you can to pass through The non-chrominance information of statistics a large amount of (at least 25 scene switching film sources) lacks image Tm_frame_u,t(2)+Tm_frame_v,t(2)

Numeric distribution, determine numerical value corresponding to maximum probability as colourity first kind decision threshold and threshold value；

Thres_u1 is the relative threshold that colourity u is corresponded to when chrominance information changes drastic scene, can pass through statistical experiment Obtain, you can to lack image by the non-chrominance information for counting a large amount of (film sources that at least 25 scenes switch)

(Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)

(Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)

Numeric distribution, determine to correspond to colourity u's when numerical value corresponding to maximum probability changes drastic scene as chrominance information Relative threshold.

Thres_v1 is the relative threshold that colourity v is corresponded to when chrominance information changes drastic scene, can pass through statistical experiment Obtain, you can to lack image by the non-chrominance information for counting a large amount of (film sources that at least 25 scenes switch)

(Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)

(Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)

Numeric distribution, determine to correspond to colourity v's when numerical value corresponding to maximum probability changes drastic scene as chrominance information Relative threshold.Non- chrominance information lacks image and refers to that the colourity energy in image at least in the presence of a pixel is more than decision threshold Thres_upImage

I.e.

WhereinFor the colourity energy of a pixel, u (i, j), v (i, j), divide Not Wei positioned at image the i-th row j row chromatic components u, v numerical value, Thres_upLack the decision threshold of image for non-chrominance information, Thres_up>30；

Else if

((Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)>Thres_u2||

(Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)>Thres_u2))

And

((Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)>Thres_v2||

(Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)>Thres_v2)

Then：(chrominance information changes the situation of small scene)

It is respectively brightness the second class decision threshold Thres_y_3 to make the careful decision threshold Thres_y1 and Thres_y2 of brightness And Thres_y_4；The careful decision threshold Thres_uv1 and Thres_uv2 of assumed appearance degree is respectively the careful decision threshold of the class of colourity second Value Thres_uv_3 and Thres_uv_4；I.e.

Thres_y1=Thres_y_3, Thres_y2=Thres_y_4

Thres_uv1=Thres_uv_3, Thres_uv2=Thres_uv_4

Into module 43,

Wherein, Thres_y_3 is brightness the second class decision threshold relative threshold, can be obtained by statistical experiment, you can To lack image by the chrominance information for counting a large amount of (at least 25 scenes switch film source)

(Ts_frame_y,t(2)-Ts_frame_y,t(1))/Ts_frame_y,t(2)、

(Ts_frame_y,t(2)-Ts_frame_y,t(3))/Ts_frame_y,t(2)

Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright the second class decision threshold relative threshold；

Thres_y_4 is brightness the second class decision threshold absolute threshold, can be obtained by statistical experiment, you can to pass through The chrominance information of statistics a large amount of (at least 25 scene switching film sources) lacks image Ts_frame_y,t(2)-Ts_frame_y,t(1)、

Ts_frame_y,t(2)-Ts_frame_y,t(3)

Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright the second class decision threshold absolute threshold；

Thres_uv_3 is colourity the second class decision threshold value difference threshold value, can be obtained by statistical experiment, you can to pass through The chrominance information of statistics a large amount of (at least 25 scene switching film sources) lacks image fabs (Tm_frame_u,t(2)-Tm_frame_v,t (2))

Numeric distribution, determine numerical value corresponding to maximum probability as colourity the second class decision threshold value difference threshold value；

Thres_uv_4 is colourity the second class decision threshold and threshold value, can be obtained by statistical experiment, you can to pass through The chrominance information of statistics a large amount of (at least 25 scene switching film sources) lacks image Tm_frame_u,t(2)+Tm_frame_v,t(2)

Numeric distribution, determine numerical value corresponding to maximum probability as colourity the second class decision threshold and threshold value；

Thres_u2 is the relative threshold that colourity u is corresponded to when chrominance information changes small scene, can be obtained by statistical experiment Take, you can to lack image by the chrominance information for counting a large amount of (film sources that at least 25 scenes switch)

(Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)

(Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)

Numeric distribution, determine to correspond to colourity u phase when numerical value corresponding to maximum probability changes small scene as chrominance information To threshold value.

Thres_v2 is the relative threshold that colourity v is corresponded to when chrominance information changes small scene, can be obtained by statistical experiment Take, you can to lack image by the chrominance information for counting a large amount of (film sources that at least 25 scenes switch)

(Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)

(Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)

Numeric distribution, when determining that numerical value corresponding to maximum probability changes small scene as chrominance information

Corresponding colourity v relative threshold.Chrominance information lacks image and refers to that image all pixels point colourity energy is respectively less than Decision threshold Thres_downImage

I.e.

WhereinFor the colourity energy of a pixel, u (i, j), v (i, j), divide Wei not be positioned at image the i-th row j row chromatic components u, v numerical value, Thres_downLack the decision threshold of image for chrominance information, Thres_down<15,

Relative threshold Thres_y_1, the brightness first kind decision threshold absolute threshold of the brightness first kind decision threshold Thres_y_2, colourity first kind decision threshold value difference threshold value Thres_uv_1, colourity first kind decision threshold and threshold value Thres_ When colourity u relative threshold Thres_u1, chrominance information change drastic scene is corresponded to when uv_2, chrominance information change drastic scene Corresponding colourity v relative threshold Thres_v1, brightness the second class decision threshold relative threshold Thres_y_3, the class of brightness second are sentenced Determine threshold value absolute threshold Thres_y_4, colourity the second class decision threshold value difference threshold value Thres_uv_3, colourity the second class decision threshold Colourity u relative threshold Thres_u2, chrominance information change is corresponded to when changing small scene with threshold value Thres_uv_4, chrominance information In the relative threshold Thres_v2 that colourity v is corresponded to during small scene, numeric distribution maximum is general used by each threshold value acquisition methods Rate method can also be substituted for averaging method, that is, useAs threshold value, whereinRepresent to sum to k, k represents statistical variable Concrete numerical value, p (k) represents the probability that numerical value k occurs.

Otherwise：

The non-scene switch frame of t frames is determined, makes t=t+1, reenters the judgement that module 41 carries out next frame.

Further, " it is used to be changed according to the statistical information of brightness and/or colourity, determines whether that scene is cut in module 43 Change " be specially：

Judge, if

((Ts_frame_y,t(2)-Ts_frame_y,t(1))/Ts_frame_y,t(2)>Thres_y1||

(Ts_frame_y,t(2)-Ts_frame_y,t(3))/Ts_frame_y,t(2)>Thres_y1)

And

((Ts_frame_y,t(2)-Ts_frame_y,t(1))>Thres_y2||

And

(fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))>Thres_uv1||

fabs(Tm_frame_u,t(2)+Tm_frame_v,t(2))>Thres_uv2)

And

(fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))+△>fabs(Tm_frame_u,t(1)-Tm_frame_v,t (1))

&&

fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))+△>fabs(Tm_frame_u,t(3)-Tm_frame_v,t (3))) then：Determine t two field pictures F_tFor the start frame of new scene.

Wherein, △ is drift value constant, can be obtained by statistical experiment, i.e., by counting a large amount of (at least 25 scenes The film source of switching) image

fabs(Tm_frame_u,t(1)-Tm_frame_v,t(1))-fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))

fabs(Tm_frame_u,t(3)-Tm_frame_v,t(3))-fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))

Numeric distribution, determine numerical value corresponding to maximum probability as drift value constant △.Drift value constant △ is asked for be adopted Numeric distribution maximum probabilistic method can also be substituted for averaging method, that is, useAs threshold value, whereinExpression pair K sums, and k represents the concrete numerical value of statistical variable, and p (k) represents the probability that numerical value k occurs.Other conventional statistic laws are applicable.

Drift value constant △ can strengthen the adaptability of algorithm, that is, strengthen simple binaryzation judge it is bad in statistical threshold When the problem of causing algorithm performance to decline.

“||”、“&&”、“fabs”："or", "AND" respectively in C language, " take absolute value computing ".

The embodiment of the present invention is by obtaining the y of two field picture to be detected, u, the time-variation-degree statistical property of v information, according to Brightness and/or the different situations of chrominance information change, determine the careful decision threshold of brightness and the careful decision threshold of colourity, according to bright The change of the statistical information of degree and/or colourity, the careful decision threshold of brightness and the careful decision threshold of colourity, carry out determining whether scene Switching.It may be such that encoder is adapted dynamically coding strategy using present invention method, so as to Optimized Coding Based performance.

Can it will be understood by those skilled in the art that realizing that all or part of step in above-described embodiment method is So that by programmed instruction related hardware, come what is completed, described program can be stored in a computer read/write memory medium, Described storage medium can be ROM, RAM, disk, CD etc..

The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention All any modification, equivalent and improvement made within refreshing and principle etc., should be included in the scope of the protection.

Claims (11)

1. A kind of 1. scene change detection method based on time-variation-degree, it is characterised in that methods described includes：

   Step A, the y of two field picture to be detected, u, the time-variation-degree statistical property of v information are obtained respectively；

   Step B, the different situations changed according to brightness and chrominance information, determines the careful decision threshold of brightness and the careful judgement of colourity Threshold value；

   Step C, according to the change of the statistical information of brightness and colourity, the careful decision threshold of brightness and the careful decision threshold of colourity, enter Row determines whether that scene switches；

   The step A is specifically included：

   Step a, determine two field picture key area Region to be detected_t；

   Step b, judges whether the current block in frame to be detected belongs to two field picture key area to be detected, is then to enter step c, no Then enter next piece of block of current block_t,n+1, return to step b and judged；

   Step c, calculate current block block_t,nThe first statistical property vector of y, u, v information time change degree, the second statistics it is special Property vector；

   Step d, judge whether all blocks have all asked for statistical property vector in current frame image key area, be to enter Step e, otherwise into next piece of block of current block_t,n+1, reenter step b；

   Step e, calculate the first statistical property, the second statistical property of y, u, v information time change degree of two field picture to be detected；

   Wherein：T represents the frame number of frame to be detected in the video sequence, and t frames are frame to be detected, Region_tRepresent t frame figures As key area, block_t,nN-th piece of t two field pictures is represented, n-th piece is to be detected piece, namely current block, block_t,n+1Table Show t two field pictures (n+1)th piece, y represent the luminance component of image, and u, v represent the chromatic component of image respectively；

   The step B is specially：

   If

   ((Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)>Thres_u1||

   (Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)>Thres_u1)

   Or

   ((Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)>Thres_v1||

   (Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)>Thres_v1)

   Then：

   Make the careful decision threshold Thres_y1 and Thres_y2 of brightness be respectively brightness first kind decision threshold Thres_y_1 and Thres_y_2；The careful decision threshold Thres_uv1 and Thres_uv2 of assumed appearance degree is respectively the careful decision threshold of the colourity first kind Thres_uv_1 and Thres_uv_2；I.e.

   Thres_y1=Thres_y_1, Thres_y2=Thres_y_2

   Thres_uv1=Thres_uv_1, Thres_uv2=Thres_uv_2

   Then, into step C

   Wherein, Thres_y_1 is the relative threshold of brightness first kind decision threshold, switches film source by counting at least 25 scenes Non- chrominance information lack image

   (Ts_frame_y,_t(2)-Ts_frame_y,_t(1))/Ts_frame_y,_t(2)、

   (Ts_frame_y,_t(2)-Ts_frame_y,_t(3))/Ts_frame_y,_t(2)

   Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright first kind decision threshold relative threshold；

   Thres_y_2 is brightness first kind decision threshold absolute threshold, switches the non-color of film source by counting at least 25 scenes Spend poor information image Ts_frame_y,t(2)-Ts_frame_y,t(1)、Ts_frame_y,t(2)-Ts_frame_y,t(3)

   Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright first kind decision threshold absolute threshold；

   Thres_uv_1 is colourity first kind decision threshold value difference threshold value, switches the non-colourity of film source by counting at least 25 scenes Poor information image fabs (Tm_frame_u,t(2)-Tm_frame_v,t(2)) numeric distribution, numerical value corresponding to maximum probability is determined As colourity first kind decision threshold value difference threshold value；

   Thres_uv_2 is colourity first kind decision threshold and threshold value, switches the non-colourity of film source by counting at least 25 scenes Poor information image Tm_frame_u,t(2)+Tm_frame_v,t(2)

   Numeric distribution, determine numerical value corresponding to maximum probability as colourity first kind decision threshold and threshold value；

   Thres_u1 is the relative threshold that colourity u is corresponded to when chrominance information changes drastic scene, by counting at least 25 scenes The non-chrominance information of the film source of switching lacks image

   (Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)

   (Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)

   Numeric distribution, determine to correspond to the relative of colourity u when numerical value corresponding to maximum probability changes drastic scene as chrominance information Threshold value,

   Thres_v1 is the relative threshold that colourity v is corresponded to when chrominance information changes drastic scene, by counting at least 25 scenes The non-chrominance information of the film source of switching lacks image

   (Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)

   (Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)

   Numeric distribution, determine to correspond to the relative of colourity v when numerical value corresponding to maximum probability changes drastic scene as chrominance information Threshold value,

   Non- chrominance information lacks image and refers to that the colourity energy in image at least in the presence of a pixel is more than decision threshold Thres_upImage

   I.e.

   WhereinFor the colourity energy of a pixel, u (i, j), v (i, j), it is respectively Positioned at image the i-th row j row chromatic components u, v numerical value, Thres_upLack the decision threshold of image for non-chrominance information, Thres_up>30；

   Else if

   ((Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)>Thres_u2||

   (Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)>Thres_u2))

   And

   ((Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)>Thres_v2||

   (Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)>Thres_v2)

   Then：

   Make the careful decision threshold Thres_y1 and Thres_y2 of brightness be respectively brightness the second class decision threshold Thres_y_3 and Thres_y_4, the careful decision threshold Thres_uv1 and Thres_uv2 of assumed appearance degree are respectively the careful decision threshold of the class of colourity second Thres_uv_3 and Thres_uv_4；I.e.

   <mrow> <mtable> <mtr> <mtd> <mrow> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>y</mi> <mn>1</mn> <mo>=</mo> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>y</mi> <mo>_</mo> <mn>3</mn> <mo>,</mo> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>y</mi> <mn>2</mn> <mo>=</mo> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>y</mi> <mo>_</mo> <mn>4</mn> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>u</mi> <mi>v</mi> <mn>1</mn> <mo>=</mo> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>u</mi> <mi>v</mi> <mo>_</mo> <mn>3</mn> <mo>,</mo> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>u</mi> <mi>v</mi> <mn>2</mn> <mo>=</mo> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>u</mi> <mi>v</mi> <mo>_</mo> <mn>4</mn> </mrow> </mtd> </mtr> </mtable> <mo>,</mo> </mrow>

   Into step C,

   Wherein, Thres_y_3 is brightness the second class decision threshold relative threshold, switches film source by counting at least 25 scenes Chrominance information lacks image

   (Ts_frame_y,t(2)-Ts_frame_y,t(1))/Ts_frame_y,t(2)、

   (Ts_frame_y,t(2)-Ts_frame_y,t(3))/Ts_frame_y,t(2)

   Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright the second class decision threshold relative threshold；

   Thres_y_4 is brightness the second class decision threshold absolute threshold, switches the colourity of film source by counting at least 25 scenes Poor information image Ts_frame_y,t(2)-Ts_frame_y,t(1)、Ts_frame_y,t(2)-Ts_frame_y,t(3)

   Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright the second class decision threshold absolute threshold；

   Thres_uv_3 is colourity the second class decision threshold value difference threshold value, and the colourity for switching film source by counting at least 25 scenes is believed Breath lacks image fabs (Tm_frame_u,t(2)-Tm_frame_v,t(2)) numeric distribution, determine that numerical value corresponding to maximum probability is made For colourity the second class decision threshold value difference threshold value；

   Thres_uv_4 is colourity the second class decision threshold and threshold value, and the colourity for switching film source by counting at least 25 scenes is believed Breath lacks image Tm_frame_u,t(2)+Tm_frame_v,t(2) numeric distribution, determine that numerical value is as colourity corresponding to maximum probability Second class decision threshold and threshold value；

   Thres_u2 is the relative threshold that colourity u is corresponded to when chrominance information changes small scene, is cut by counting at least 25 scenes The chrominance information for the film source changed lacks image

   (Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)

   (Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)

   Numeric distribution, determine to correspond to colourity u relative threshold when numerical value corresponding to maximum probability changes small scene as chrominance information Value,

   Thres_v2 is the relative threshold that colourity v is corresponded to when chrominance information changes small scene, is cut by counting at least 25 scenes The chrominance information for the film source changed lacks image

   (Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)

   (Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)

   Numeric distribution, when determining that numerical value corresponding to maximum probability changes small scene as chrominance information

   Corresponding colourity v relative threshold,

   Chrominance information lacks image and refers to image all pixels point color

   Degree energy is respectively less than decision threshold Thres_downImage

   I.e.

   WhereinFor the colourity energy of a pixel, u (i, j), v (i, j), it is respectively Positioned at image the i-th row j row chromatic components u, v numerical value, Thres_downLack the decision threshold of image for chrominance information, Thres_down<15,

   Otherwise：

   The non-scene switch frame of t frames is determined, makes t=t+1, reenters the judgement that the step A enters next frame；

   Ts_frame_y,tReferred to as the first statistical property of t two field pictures monochrome information time-variation-degree,

   Ts_frame_u,t, Ts_frame_v,tReferred to as the first statistical property of t two field pictures chrominance information time-variation-degree,

   Tm_frame_y,tReferred to as the second statistical property of t two field pictures monochrome information time-variation-degree,

   Tm_frame_u,t, Tm_frame_v,tReferred to as the second statistical property of t two field pictures chrominance information time-variation-degree；

   Ts_frame_y,t(1)、Ts_frame_y,t(2)、Ts_frame_y,t(3) it is respectively Ts_frame_y,tThe representation of each component；

   Ts_frame_u,t(1)、Ts_frame_u,t(2)Ts_frame_u,t(3) it is respectively Ts_frame_u,tThe representation of each component；

   Ts_frame_v,t(1)、Ts_frame_v,t(2)、Ts_frame_v,t(3) it is respectively Ts_frame_v,tThe representation of each component；

   Tm_frame_y,t(1)、Tm_frame_y,t(2)、Tm_frame_y,t(3) it is respectively Tm_frame_y,tThe representation of each component；

   Tm_frame_u,t(1)、Tm_frame_u,t(2)、Tm_frame_u,t(3) it is respectively Tm_frame_u,tThe representation of each component；

   Tm_frame_v,t(1)、Tm_frame_v,t(2)、Tm_frame_v,t(3) it is respectively Tm_frame_v,tThe representation of each component；

   The step C is specially：

   If

   ((Ts_frame_y,t(2)-Ts_frame_y,t(1))/Ts_frame_y,t(2)>Thres_y1||

   (Ts_frame_y,t(2)-Ts_frame_y,t(3))/Ts_frame_y,t(2)>Thres_y1)

   And

   ((Ts_frame_y,t(2)-Ts_frame_y,t(1))>Thres_y2||

   (Ts_frame_y,t(2)-Ts_frame_y,t(3))>Thres_y2)

   And

   (fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))>Thres_uv1||

   fabs(Tm_frame_u,t(2)+Tm_frame_v,t(2))>Thres_uv2)

   And

   (fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))+△>fabs(Tm_frame_u,t(1)-Tm_frame_v,t(1))

   &&

   fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))+△>fabs(Tm_frame_u,t(3)-Tm_frame_v,t(3)))

   Then：The start frame that t two field pictures are new scene is determined,

   Wherein, △ is drift value constant, " | | ", " ＆＆ ", " fabs " be respectively "or" in C language, "AND", " take absolute value fortune Calculate ", △ passes through the statistics piece source images that at least 25 scenes switch

   fabs(Tm_frame_u,t(1)-Tm_frame_v,t(1))-fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2)) and

   fabs(Tm_frame_u,t(3)-Tm_frame_v,t(3))-fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))

   Numeric distribution, determine numerical value corresponding to maximum probability as drift value constant △；

   Wherein, the step c is specially：

   c1：Obtain block_t,nF information time change degrees 3 set b_f,t-2,n、b_f,t-1,n、b_f,t,n；

   F is respectively equal to y, u, v, and y represents the luminance component of image, and u, v represent the chromatic component of image, set b respectively_f,t-2,n、 b_f,t-1,n、b_f,t,nCalculation formula such as shown in (1)：

   <mrow> <msub> <mi>b</mi> <mrow> <mi>f</mi> <mo>,</mo> <mi>m</mi> <mo>,</mo> <mi>n</mi> </mrow> </msub> <mo>=</mo> <mo>{</mo> <munder> <mrow> <msub> <mi>f</mi> <mrow> <mi>m</mi> <mo>+</mo> <mn>1</mn> </mrow> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> <mo>-</mo> <msub> <mi>f</mi> <mi>m</mi> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> </mrow> <mrow> <msub> <mi>f</mi> <mrow> <mi>m</mi> <mo>+</mo> <mn>1</mn> </mrow> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> <mo>&amp;Element;</mo> <msub> <mi>block</mi> <mrow> <mi>m</mi> <mo>+</mo> <mn>1</mn> <mo>,</mo> <mi>n</mi> </mrow> </msub> <mo>&amp;cap;</mo> <msub> <mi>f</mi> <mi>m</mi> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> <mo>&amp;Element;</mo> <msub> <mi>block</mi> <mrow> <mi>m</mi> <mo>,</mo> <mi>n</mi> </mrow> </msub> </mrow> </munder> <mo>}</mo> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow>

   M in formula (1) is respectively equal to t-2, t-1, t, you can obtains b respectively_f,t-2,n、b_f,t-1,n、b_f,t,n, block_m,nRepresent the N-th piece of m two field pictures,

   block_m+1,nN-th piece of m+1 two field pictures are represented,

   f_m(i, j) represents the numerical value of m two field picture the i-th row jth row f information,

   f_m+1,n(i, j) represents the numerical value of m+1 two field picture the i-th row jth row f information,

   f_m(i,j)∈block_m,nExpression is located at block_m,nThe numerical value of m two field pictures the i-th row jth row f information in block,

   f_m+1(i,j)∈block_m+1,nExpression is located at block_m+1,nThe numerical value of m+1 two field pictures the i-th row jth row f information in block,

   Expression meets f_m+1(i,j)∈block_m+1,nAnd f_m(i,j)∈block_m,nAll f_m+1 (i,j)-f_mThe set of (i, j),

   f_m+1(i,j)-f_m(i, j) is the subtraction of the numerical value of corresponding f information,

   c2：Ask for block_t,n3 set b of f information times change degree_f,t-2,n、b_f,t-1,n、b_f,t,nThe first statistical property, point Std (b are not designated as_f,t-2,n)、Std(b_f,t-1,n)、Std(b_f,t,n), Std represents to seek mean square deviation；

   c3：Ask for block_t,n3 set b of f information times change degree_f,t-2,n、b_f,t-1,n、b_f,t,nThe second statistical property, point Mean (b are not designated as_f,t-2,n)、mean(b_f,t-1,n)、mean(b_f,t,n), mean represents to average；

   c4：Build block_t,nF information times change degree the first statistical property vector T s_b_f,t,nWith the second statistical property vector Tm_b_f,t,n, its construction method is as follows：

   Ts_b_f,t,n=(Std (b_f,t-2,n),Std(b_f,t-1,n),Std(b_f,t,n)) (2)

   Tm_b_f,t,n=(mean (b_f,t-2,n),mean(b_f,t-1,n),mean(b_f,t,n)) (3)；

   Wherein, the step e is specially：

   To all pieces in t two field picture key areas of Ts_b_f,t,nAverage, the f information time change degrees as t two field pictures First statistical property Ts_frame_f,t, to all pieces in t two field picture key areas of Tm_b_f,t,nAverage, as t two field pictures F information time change degrees the second statistical property Tm_frame_f,t,

   Ts_frame_f,t=mean (Ts_b_f,t,n) (4)

   Tm_frame_f,t=mean (Tm_b_f,t,n) (5)。
2. 2. the scene change detection method based on time-variation-degree as claimed in claim 1, it is characterised in that the step A Also include step before：

   Down-sampling is carried out to pending original image.
3. 3. the scene change detection method based on time-variation-degree as claimed in claim 1, it is characterised in that the step a Specially：

   Key area Region using whole t two field pictures as t two field pictures_t, or

   The central area of t two field pictures is taken as t two field picture key areas, or

   In the case where ensureing certain accuracy of detection, the key area of image is determined by reducing determinating area.
4. 4. the scene change detection method based on time-variation-degree as claimed in claim 3, it is characterised in that described " to take t The central area of two field picture is as t two field pictures key area " be specially：

   Remove the height/k of t two field pictures the top_hIndividual macro-block line, the height/k of bottom_hIndividual macro-block line, remove t again The width/k of the two field picture leftmost side_wIndividual macro block row, rightmost side width/k_wAfter individual macro-block line, remaining image is as Region_t,

   Wherein height, width is respectively the line number of image pixel, columns, k_h、k_wRespectively line direction and column direction ratio system Number, the proportionality coefficient is integer.
5. 5. the scene change detection method based on time-variation-degree as claimed in claim 1, it is characterised in that

   Utilize formula (2), (3), formula (4), (5), the second statistical property Ts_ of the f information time change degrees of t two field pictures frame_f,tWith the second statistical property Tm_frame_f,tFurther it is embodied as：

   <mrow> <mtable> <mtr> <mtd> <mrow> <mi>T</mi> <mi>s</mi> <mo>_</mo> <msub> <mi>frame</mi> <mrow> <mi>f</mi> <mo>,</mo> <mi>t</mi> </mrow> </msub> <mo>=</mo> <mi>m</mi> <mi>e</mi> <mi>a</mi> <mi>n</mi> <mrow> <mo>(</mo> <mo>(</mo> <mrow> <mi>S</mi> <mi>t</mi> <mi>d</mi> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mrow> <mi>f</mi> <mo>,</mo> <mi>t</mi> <mo>-</mo> <mn>2</mn> <mo>,</mo> <mi>n</mi> </mrow> </msub> <mo>)</mo> </mrow> <mo>,</mo> <mi>S</mi> <mi>t</mi> <mi>d</mi> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mrow> <mi>f</mi> <mo>,</mo> <mi>t</mi> <mo>-</mo> <mn>1</mn> <mo>,</mo> <mi>n</mi> </mrow> </msub> <mo>)</mo> </mrow> <mo>,</mo> <mi>S</mi> <mi>t</mi> <mi>d</mi> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mrow> <mi>f</mi> <mo>,</mo> <mi>t</mi> <mo>,</mo> <mi>n</mi> </mrow> </msub> <mo>)</mo> </mrow> </mrow> <mo>)</mo> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mo>=</mo> <mrow> <mo>(</mo> <mi>m</mi> <mi>e</mi> <mi>a</mi> <mi>n</mi> <mo>(</mo> <mrow> <mi>S</mi> <mi>t</mi> <mi>d</mi> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mrow> <mi>f</mi> <mo>,</mo> <mi>t</mi> <mo>-</mo> <mn>2</mn> <mo>,</mo> <mi>n</mi> </mrow> </msub> <mo>)</mo> </mrow> </mrow> <mo>)</mo> <mo>,</mo> <mi>m</mi> <mi>e</mi> <mi>a</mi> <mi>n</mi> <mo>(</mo> <mrow> <mi>S</mi> <mi>t</mi> <mi>d</mi> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mrow> <mi>f</mi> <mo>,</mo> <mi>t</mi> <mo>-</mo> <mn>1</mn> <mo>,</mo> <mi>n</mi> </mrow> </msub> <mo>)</mo> </mrow> </mrow> <mo>)</mo> <mo>,</mo> <mi>m</mi> <mi>e</mi> <mi>a</mi> <mi>n</mi> <mo>(</mo> <mrow> <mi>S</mi> <mi>t</mi> <mi>d</mi> <mrow> <mo>(</mo> <msub> <mi>b</mi> <mrow> <mi>f</mi> <mo>,</mo> <mi>t</mi> <mo>,</mo> <mi>n</mi> </mrow> </msub> <mo>)</mo> </mrow> </mrow> <mo>)</mo> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> </mtable> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>6</mn> <mo>)</mo> </mrow> </mrow>

   Tm_frame_f,t=mean ((mean (b_f,t-2,n),mean(b_f,t-1,n),mean(b_f,t,n)))

   =(mean (mean (b_f,t-2,n)),mean(mean(b_f,t-1,n)),mean(mean(b_f,t,n))) (7)。
6. 6. the scene change detection method based on time-variation-degree as claimed in claim 5, it is characterised in that

   Each component of vector is represented using sequence number of the component of vector in vector,

   Ts_b_f,t,nEach representation in components be：

   Ts_b_f,t,n(1)=Std (b_f,t-2,n), Ts_b_f,t,n(2)=Std (b_f,t-1,n), Ts_b_f,t,n(3)=Std (b_f,t,n),

   Tm_b_f,t,nEach representation in components be：

   Tm_b_f,t,n(1)=mean (b_f,t-2,n), Tm_b_f,t,n(2)=mean (b_f,t-1,n), Tm_b_f,t,n(3)=mean (b_f,t,n)；

   Ts_frame_f,tEach representation in components be：

   Ts_frame_f,t(1)=mean (Std (b_f,t-2,n)),

   Ts_frame_f,t(2)=mean (Std (b_f,t-1,n)),

   Ts_frame_f,t(3)=mean (Std (b_f,t,n)), (8)

   Tm_frame_f,tEach component be represented by：

   Tm_frame_f,t(1)=mean (mean (b_f,t-2,n)),

   Tm_frame_f,t(2)=mean (mean (b_f,t-1,n)),

   Tm_frame_f,t(3)=mean (mean (b_f,t,n)), (9)

   Make f in (6) be respectively equal to y, u, v, ask for Ts_frame_y,t, Ts_frame_u,t, Ts_frame_v,t,

   Make f in (7) be respectively equal to y, u, v, ask for Tm_frame_y,t, Tm_frame_u,t, Tm_frame_v,t；

   Ts_frame_y,tReferred to as the first statistical property of t two field pictures monochrome information time-variation-degree,

   Ts_frame_u,t, Ts_frame_v,tReferred to as the first statistical property of t two field pictures chrominance information time-variation-degree,

   Tm_frame_y,tReferred to as the second statistical property of t two field pictures monochrome information time-variation-degree,

   Tm_frame_u,t, Tm_frame_v,tReferred to as the second statistical property of t two field pictures chrominance information time-variation-degree.
7. 7. the scene change detection method based on time-variation-degree as claimed in claim 6, it is characterised in that utilize vector Sequence number of the component in vector represents each component of vector,

   Make f in (8) be respectively equal to y, u, v, obtain Ts_frame_y,t、Ts_frame_u,t、Ts_frame_v,tThe representation of each component；

   Make f in (9) be respectively equal to y, u, v, obtain Tm_frame_y,t、Tm_frame_u,t、Tm_frame_v,tThe representation of each component；

   Ts_frame_y,tEach component be represented by：

   Ts_frame_y,t(1)=mean (Std (b_y,t-2,n))；

   Ts_frame_y,t(2)=mean (Std (b_y,t-1,n))；

   Ts_frame_y,t(3)=mean (Std (b_y,t,n))；

   Ts_frame_u,tEach component be represented by：

   Ts_frame_u,t(1)=mean (Std (b_u,t-2,n))；

   Ts_frame_u,t(2)=mean (Std (b_u,t-1,n))；

   Ts_frame_u,t(3)=mean (Std (b_u,t,n))；

   Ts_frame_v,tEach component be represented by：

   Ts_frame_v,t(1)=mean (Std (b_v,t-2,n))；

   Ts_frame_v,t(2)=mean (Std (b_v,t-1,n))；

   Ts_frame_v,t(3)=mean (Std (b_v,t,n))；

   Tm_frame_y,tEach component be represented by：

   Tm_frame_y,t(1)=mean (mean (b_y,t-2,n))；

   Tm_frame_y,t(2)=mean (mean (b_y,t-1,n))；

   Tm_frame_y,t(3)=mean (mean (b_y,t,n))；

   Tm_frame_u,tEach component be represented by：

   Tm_frame_u,t(1)=mean (mean (b_u,t-2,n))；

   Tm_frame_u,t(2)=mean (mean (b_u,t-1,n))；

   Tm_frame_u,t(3)=mean (mean (b_u,t,n))；

   Tm_frame_v,tEach component be represented by：

   Tm_frame_v,t(1)=mean (mean (b_v,t-2,n))；

   Tm_frame_v,t(2)=mean (mean (b_v,t-1,n))；

   Tm_frame_v,t(3)=mean (mean (b_v,t,n))。
8. 8. the scene change detection method based on time-variation-degree as claimed in claim 1, it is characterised in that the brightness The relative threshold Thres_y_1 of a kind of decision threshold, brightness first kind decision threshold absolute threshold Thres_y_2, colourity first Class decision threshold value difference threshold value Thres_uv_1, colourity first kind decision threshold and threshold value Thres_uv_2, chrominance information change are acute Colourity v relative threshold is corresponded to when relative threshold Thres_u1, chrominance information change drastic scene that colourity u is corresponded to during strong scene Thres_v1, brightness the second class decision threshold relative threshold Thres_y_3, brightness the second class decision threshold absolute threshold Thres_ Y_4, colourity the second class decision threshold value difference threshold value Thres_uv_3, colourity the second class decision threshold and threshold value Thres_uv_4, color Correspond to colourity v's when the relative threshold Thres_u2, the chrominance information small scene of change that correspond to colourity u during degree information change small scene In relative threshold Thres_v2, numeric distribution maximum probabilistic method can also be substituted for average used by each threshold value acquisition methods Method, that is, useAs threshold value, whereinRepresent to sum to k, k represents the concrete numerical value of statistical variable, and p (k) represents number The probability that value k occurs.
9. 9. the scene change detection method based on time-variation-degree as claimed in claim 1, it is characterised in that ask for drift value Numeric distribution maximum probabilistic method can also be substituted for averaging method used by constant △, that is, useAs threshold value, its InRepresent to sum to k, k represents the concrete numerical value of statistical variable, and p (k) represents the probability that numerical value k occurs.
10. 10. a kind of scene change detection device based on time-variation-degree, it is characterised in that described device includes：Two field picture y u The time-variation-degree statistical property acquisition module (41) of v information, careful decision threshold acquisition module (4) 2, scene are switched and determined mould Block (43),

    The time-variation-degree statistical property acquisition module (41) of two field picture y u v information, for obtaining two field picture to be detected respectively F information time-variation-degree statistical property, f is respectively equal to y, u, v, and y represents the luminance component of image, and u, v represent to scheme respectively The chromatic component of picture；

    Careful decision threshold acquisition module (42), for the different situations according to brightness and chrominance information change, determine that brightness is thin Cause decision threshold and the careful decision threshold of colourity；

    Specially：

    If

    ((Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)>Thres_u1||

    (Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)>Thres_u1)

    Or

    ((Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)>Thres_v1||

    (Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)>Thres_v1)

    Then：

    Make the careful decision threshold Thres_y1 and Thres_y2 of brightness be respectively brightness first kind decision threshold Thres_y_1 and Thres_y_2；The careful decision threshold Thres_uv1 and Thres_uv2 of assumed appearance degree is respectively the careful decision threshold of the colourity first kind Thres_uv_1 and Thres_uv_2；I.e.

    Thres_y1=Thres_y_1, Thres_y2=Thres_y_2

    Thres_uv1=Thres_uv_1, Thres_uv2=Thres_uv_2

    Then, module (43) is switched and determined into scene,

    Wherein, Thres_y_1 is the relative threshold of brightness first kind decision threshold, switches film source by counting at least 25 scenes Non- chrominance information lack image

    (Ts_frame_y,t(2)-Ts_frame_y,t(1))/Ts_frame_y,t(2)、

    (Ts_frame_y,t(2)-Ts_frame_y,t(3))/Ts_frame_y,t(2)

    Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright first kind decision threshold relative threshold；

    Thres_y_2 is brightness first kind decision threshold absolute threshold, switches the non-color of film source by counting at least 25 scenes Spend poor information image Ts_frame_y,t(2)-Ts_frame_y,t(1)、Ts_frame_y,t(2)-Ts_frame_y,t(3)

    Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright first kind decision threshold absolute threshold；

    Thres_uv_1 is colourity first kind decision threshold value difference threshold value, switches the non-colourity of film source by counting at least 25 scenes Poor information image fabs (Tm_frame_u,t(2)-Tm_frame_v,t(2))

    Numeric distribution, determine numerical value corresponding to maximum probability as colourity first kind decision threshold value difference threshold value；

    Thres_uv_2 is colourity first kind decision threshold and threshold value, switches the non-colourity of film source by counting at least 25 scenes Poor information image Tm_frame_u,t(2)+Tm_frame_v,t(2)

    Numeric distribution, determine numerical value corresponding to maximum probability as colourity first kind decision threshold and threshold value；

    Thres_u1 is the relative threshold that colourity u is corresponded to when chrominance information changes drastic scene, by counting at least 25 scenes The non-chrominance information of the film source of switching lacks image

    (Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)

    (Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)

    Numeric distribution, determine to correspond to the relative of colourity u when numerical value corresponding to maximum probability changes drastic scene as chrominance information Threshold value,

    Thres_v1 is the relative threshold that colourity v is corresponded to when chrominance information changes drastic scene, by counting at least 25 scenes The non-chrominance information of the film source of switching lacks image

    (Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)

    (Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)

    Numeric distribution, determine to correspond to the relative of colourity v when numerical value corresponding to maximum probability changes drastic scene as chrominance information Threshold value,

    Non- chrominance information lacks image and refers to that the colourity energy in image at least in the presence of a pixel is more than decision threshold Thres_upImage

    I.e.

    WhereinFor the colourity energy of a pixel, u (i, j), v (i, j), it is respectively Positioned at image the i-th row j row chromatic components u, v numerical value, Thres_upLack the decision threshold of image for non-chrominance information, Thres_up>30；

    Else if

    ((Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)>Thres_u2||

    (Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)>Thres_u2))

    And

    ((Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)>Thres_v2||

    (Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)>Thres_v2)

    Then：

    Make the careful decision threshold Thres_y1 and Thres_y2 of brightness be respectively brightness the second class decision threshold Thres_y_3 and Thres_y_4, the careful decision threshold Thres_uv1 and Thres_uv2 of assumed appearance degree are respectively the careful decision threshold of the class of colourity second Thres_uv_3 and Thres_uv_4；I.e.

    <mrow> <mtable> <mtr> <mtd> <mrow> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>y</mi> <mn>1</mn> <mo>=</mo> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>y</mi> <mo>_</mo> <mn>3</mn> <mo>,</mo> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>y</mi> <mn>2</mn> <mo>=</mo> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>y</mi> <mo>_</mo> <mn>4</mn> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>u</mi> <mi>v</mi> <mn>1</mn> <mo>=</mo> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>u</mi> <mi>v</mi> <mo>_</mo> <mn>3</mn> <mo>,</mo> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>u</mi> <mi>v</mi> <mn>2</mn> <mo>=</mo> <mi>T</mi> <mi>h</mi> <mi>r</mi> <mi>e</mi> <mi>s</mi> <mo>_</mo> <mi>u</mi> <mi>v</mi> <mo>_</mo> <mn>4</mn> </mrow> </mtd> </mtr> </mtable> <mo>,</mo> </mrow>

    Module (43) is switched and determined into scene,

    Wherein, Thres_y_3 is brightness the second class decision threshold relative threshold, switches film source by counting at least 25 scenes Chrominance information lacks image

    (Ts_frame_y,t(2)-Ts_frame_y,t(1))/Ts_frame_y,t(2)、

    (Ts_frame_y,t(2)-Ts_frame_y,t(3))/Ts_frame_y,t(2)

    Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright the second class decision threshold relative threshold；

    Thres_y_4 is brightness the second class decision threshold absolute threshold, switches the colourity of film source by counting at least 25 scenes Poor information image Ts_frame_y,t(2)-Ts_frame_y,t(1)、Ts_frame_y,_t(2)-Ts_frame_y,_t(3)

    Numeric distribution, determine numerical value corresponding to maximum probability as corresponding bright the second class decision threshold absolute threshold；

    Thres_uv_3 is colourity the second class decision threshold value difference threshold value, and the colourity for switching film source by counting at least 25 scenes is believed Breath lacks image fabs (Tm_frame_u,t(2)-Tm_frame_v,t(2)) numeric distribution, determine that numerical value corresponding to maximum probability is made For colourity the second class decision threshold value difference threshold value；

    Thres_uv_4 is colourity the second class decision threshold and threshold value, and the colourity for switching film source by counting at least 25 scenes is believed Breath lacks image Tm_frame_u,t(2)+Tm_frame_v,t(2)

    Numeric distribution, determine numerical value corresponding to maximum probability as colourity the second class decision threshold and threshold value；

    Thres_u2 is the relative threshold that colourity u is corresponded to when chrominance information changes small scene, is cut by counting at least 25 scenes The chrominance information for the film source changed lacks image

    (Ts_frame_u,t(2)-Ts_frame_u,t(1))/Ts_frame_u,t(2)

    (Ts_frame_u,t(2)-Ts_frame_u,t(3))/Ts_frame_u,t(2)

    Numeric distribution, determine to correspond to colourity u relative threshold when numerical value corresponding to maximum probability changes small scene as chrominance information Value,

    Thres_v2 is the relative threshold that colourity v is corresponded to when chrominance information changes small scene, is cut by counting at least 25 scenes The chrominance information for the film source changed lacks image

    (Ts_frame_v,t(2)-Ts_frame_v,t(1))/Ts_frame_v,t(2)

    (Ts_frame_v,t(2)-Ts_frame_v,t(3))/Ts_frame_v,t(2)

    Numeric distribution, determine to correspond to colourity v relative threshold when numerical value corresponding to maximum probability changes small scene as chrominance information Value,

    Chrominance information lacks image and refers to image all pixels point color

    Degree energy is respectively less than decision threshold Thres_downImage

    I.e.

    WhereinFor the colourity energy of a pixel, u (i, j), v (i, j), it is respectively Positioned at image the i-th row j row chromatic components u, v numerical value, Thres_downLack the decision threshold of image for chrominance information, Thres_down<15,

    Otherwise：

    The non-scene switch frame of t frames is determined, makes t=t+1, the time-variation-degree statistics for reentering two field picture y u v information is special Property acquisition module (41) enter next frame judgement；

    Ts_frame_y,tReferred to as the first statistical property of t two field pictures monochrome information time-variation-degree,

    Ts_frame_u,t, Ts_frame_v,tReferred to as the first statistical property of t two field pictures chrominance information time-variation-degree,

    Tm_frame_y,tReferred to as the second statistical property of t two field pictures monochrome information time-variation-degree,

    Tm_frame_u,t, Tm_frame_v,tReferred to as the second statistical property of t two field pictures chrominance information time-variation-degree；

    Ts_frame_y,t(1)、Ts_frame_y,t(2)、Ts_frame_y,t(3) it is respectively Ts_frame_y,tThe representation of each component；

    Ts_frame_u,t(1)、Ts_frame_u,t(2)Ts_frame_u,t(3) it is respectively Ts_frame_u,tThe representation of each component；

    Ts_frame_v,t(1)、Ts_frame_v,t(2)、Ts_frame_v,t(3) it is respectively Ts_frame_v,tThe representation of each component；

    Tm_frame_y,t(1)、Tm_frame_y,t(2)、Tm_frame_y,t(3) it is respectively Tm_frame_y,tThe representation of each component；

    Tm_frame_u,t(1)、Tm_frame_u,t(2)、Tm_frame_u,t(3) it is respectively Tm_frame_u,tThe representation of each component；

    Tm_frame_v,t(1)、Tm_frame_v,t(2)、Tm_frame_v,t(3) it is respectively Tm_frame_v,tThe representation of each component；

    Scene be switched and determined module (43), for according to the statistical information of brightness and colourity change, the careful decision threshold of brightness and The careful decision threshold of colourity, determine whether that scene switches；

    Specially：

    If

    ((Ts_frame_y,t(2)-Ts_frame_y,t(1))/Ts_frame_y,t(2)>Thres_y1||

    (Ts_frame_y,t(2)-Ts_frame_y,t(3))/Ts_frame_y,t(2)>Thres_y1)

    And

    ((Ts_frame_y,t(2)-Ts_frame_y,t(1))>Thres_y2||

    (Ts_frame_y,t(2)-Ts_frame_y,t(3))>Thres_y2)

    And

    (fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))>Thres_uv1||

    fabs(Tm_frame_u,t(2)+Tm_frame_v,t(2))>Thres_uv2)

    And

    (fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))+△>fabs(Tm_frame_u,t(1)-Tm_frame_v,t(1))

    &&

    fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))+△>fabs(Tm_frame_u,t(3)-Tm_frame_v,t(3)))

    Then：The start frame that t two field pictures are new scene is determined,

    Wherein, △ is drift value constant, " | | ", " ＆＆ ", " fabs " be respectively "or" in C language, "AND", " take absolute value fortune Calculate ", △ passes through the statistics piece source images that at least 25 scenes switch

    fabs(Tm_frame_u,t(1)-Tm_frame_v,t(1))-fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2)) and

    fabs(Tm_frame_u,t(3)-Tm_frame_v,t(3))-fabs(Tm_frame_u,t(2)-Tm_frame_v,t(2))

    Numeric distribution, determine numerical value corresponding to maximum probability as drift value constant △；

    The time-variation-degree statistical property acquisition module (41) of the two field picture y u v information also includes：Two field picture key area Acquisition module (411), the first judge module (412), block statistical property vector calculation module (413), the second judge module (414) the time-variation-degree statistical property computing module (415) of, two field picture y, u, v information,

    Two field picture key area acquisition module (411), for determining the image key area of frame to be detected；

    Whether the first judge module (412), the current block for judging in frame to be detected belong to two field picture key area to be detected, It is then to enter block statistical property vector calculation module (413), otherwise into next piece of current block, returns to the first judge module (412) judged；T represents the frame number of frame to be detected in the video sequence, if t frames are frame to be detected；

    Block statistical property vector calculation module (413), for calculating current block block_t,nY, u, v information time change degree First statistical property vector, the second statistical property vector；

    Specially：

    c1：Obtain block_t,nF information time change degrees 3 set b_f,t-2,n、b_f,t-1,n、b_f,t,n；

    F is respectively equal to y, u, v, and y represents the luminance component of image, and u, v represent the chromatic component of image, set b respectively_f,t-2,n、 b_f,t-1,n、b_f,t,nCalculation formula such as shown in (1)：

    <mrow> <msub> <mi>b</mi> <mrow> <mi>f</mi> <mo>,</mo> <mi>m</mi> <mo>,</mo> <mi>n</mi> </mrow> </msub> <mo>=</mo> <mo>{</mo> <munder> <mrow> <msub> <mi>f</mi> <mrow> <mi>m</mi> <mo>+</mo> <mn>1</mn> </mrow> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> <mo>-</mo> <msub> <mi>f</mi> <mi>m</mi> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> </mrow> <mrow> <msub> <mi>f</mi> <mrow> <mi>m</mi> <mo>+</mo> <mn>1</mn> </mrow> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> <mo>&amp;Element;</mo> <msub> <mi>block</mi> <mrow> <mi>m</mi> <mo>+</mo> <mn>1</mn> <mo>,</mo> <mi>n</mi> </mrow> </msub> <mo>&amp;cap;</mo> <msub> <mi>f</mi> <mi>m</mi> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> <mo>&amp;Element;</mo> <msub> <mi>block</mi> <mrow> <mi>m</mi> <mo>,</mo> <mi>n</mi> </mrow> </msub> </mrow> </munder> <mo>}</mo> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow>

    M in formula (1) is respectively equal to t-2, t-1, t, you can obtains b respectively_f,t-2,n、b_f,t-1,n、b_f,t,n, block_m,nRepresent the N-th piece of m two field pictures,

    block_m+1,nN-th piece of m+1 two field pictures are represented,

    f_m(i, j) represents the numerical value of m two field picture the i-th row jth row f information,

    f_m+1,n(i, j) represents the numerical value of m+1 two field picture the i-th row jth row f information,

    f_m(i,j)∈block_m,nExpression is located at block_m,nThe numerical value of m two field pictures the i-th row jth row f information in block,

    f_m+1(i,j)∈block_m+1,nExpression is located at block_m+1,nThe numerical value of m+1 two field pictures the i-th row jth row f information in block,

    Expression meets f_m+1(i,j)∈block_m+1,nAnd f_m(i,j)∈block_m,nAll f_m+1 (i,j)-f_mThe set of (i, j),

    f_m+1(i,j)-f_m(i, j) is the subtraction of the numerical value of corresponding f information,

    c2：Ask for block_t,n3 set b of f information times change degree_f,t-2,n、b_f,t-1,n、b_f,t,nThe first statistical property, point Std (b are not designated as_f,t-2,n)、Std(b_f,t-1,n)、Std(b_f,t,n), Std represents to seek mean square deviation；

    c3：Ask for block_t,n3 set b of f information times change degree_f,t-2,n、b_f,t-1,n、b_f,t,nThe second statistical property, point Mean (b are not designated as_f,t-2,n)、mean(b_f,t-1,n)、mean(b_f,t,n), mean represents to average；

    c4：Build block_t,nF information times change degree the first statistical property vector T s_b_f,t,nWith the second statistical property vector Tm_b_f,t,n, its construction method is as follows：

    Ts_b_f,t,n=(Std (b_f,t-2,n),Std(b_f,t-1,n),Std(b_f,t,n)) (2)

    Tm_b_f,t,n=(mean (b_f,t-2,n),mean(b_f,t-1,n),mean(b_f,t,n)) (3)；

    Described piece of statistical property vector calculation module (413) also includes：The f information time change degree set determining modules of block (4131), the f information times change of the first statistical property acquisition module (4132) of the f information time change degree set of block, block Spend the second statistical property acquisition module (4133), the first and second statistical property of the f information times change degree vector of block of set Module (4134) is built,

    The f information time change degree set determining modules (4131) of block, for obtaining to be detected piece of f information in frame to be detected 3 set of time-variation-degree, wherein f are respectively equal to y, u, v, and y represents the luminance component of image, and u, v represent image respectively Chromatic component；

    First statistical property acquisition module (4132) of the f information time change degree set of block, for asking in frame to be detected First statistical property of to be detected piece of 3 set of f information times change degree；

    Second statistical property acquisition module (4133) of the f information time change degree set of block, for asking in frame to be detected Second statistical property of to be detected piece of 3 set of f information times change degree；

    The first and second statistical property of f information times change degree vector structure module (4134) of block, for building frame to be detected In to be detected piece of the second statistical property of f information times change degree the first statistical property vector sum vector；

    Second judge module (414), for judging whether all blocks have all asked for statistics spy in t two field picture key areas Property vector, be then enter two field picture y, u, the time-variation-degree statistical property computing module (415) of v information, otherwise enter currently Next piece of block, return to the first judge module (412)；

    The time-variation-degree statistical property computing module (415) of two field picture y, u, v information, for owning according in image key area Y, u, v information time change degree statistical property vector of block, calculate the of y, u, v information time change degree of two field picture to be detected One statistical property, the second statistical property；

    It is described " to calculate two field picture to be detected in the time-variation-degree statistical property computing module (415) of two field picture y, u, v information Y, u, v information time change degree the first statistical property, the second statistical property "

    Specially：

    To all pieces in t two field picture key areas of f information times change degree the first statistical property vector T s_b_f,t,nAverage, The first statistical property Ts_frame as the f information time change degrees of t two field pictures_f,t；To institute in t two field picture key areas There is f information times change degree the second statistical property vector T m_b of block_f,t,nAverage, the f information times as t two field pictures become Second statistical property Tm_frame of change degree_f,t。
11. A kind of 11. equipment for including the scene change detection device based on time-variation-degree as claimed in claim 10.